The numbers come first: Anthropic cofounder and CEO Dario Amodei took to X on Saturday — an unusual move for an executive who generally avoids social media — to address what he called a fundamental breakdown in public confidence around artificial intelligence.
'I think it is fundamentally a crisis of trust,' Amodei wrote. 'I think that ordinary people don't trust companies, governments, or the tech industry and always suspect that we are cooking up some new way to screw them over.'
The admission is notable coming from the head of one of the most heavily funded AI labs in the world. Amodei did not place the blame on himself or any other individual AI executive, but he did not deflect the underlying charge either.
The Regulation Debate
Amodei used the post to push back on what he described as a false binary in the AI policy debate: regulatory capture on one side versus unrestricted distribution of open-weight models on the other. Neither extreme, he argued, adequately addresses the technology's risks.
He acknowledged that AI is 'structurally a technology that tends to concentrate power,' but attributed that tendency not to regulation but to AI scaling laws — the dynamic by which a model's performance improves as the resources used to build it increase. Open-weight models, he argued, merely shift that concentration toward whoever controls the most computing capacity and chips.
His preferred path: a set of 'rules of the road' that simultaneously address cyber, biological, and alignment risks; constrain the power of frontier AI companies through institutional checks; and leave room for open-weight models while managing their specific dangers. He said he supports creation of a FINRA-like regulatory entity and endorsed the Trump administration's stance on AI testing.
Promises vs. Results
On the question of negative public perception, Amodei denied that his own rhetoric is the primary driver, pointing to essays he says present both risks and benefits in equal measure. He argued that social media clips selectively amplify gloomy statements to generate clicks.
Still, he did not shy away from the harder truth. 'By far the most accurate criticism of AI companies including Anthropic is that we haven't yet delivered on our big promises to benefit the world,' he wrote.
He said Anthropic is working to change that by focusing on biology and medicine, and teased 'early glimmers' of significant results. But he vowed to avoid empty promises in the meantime, arguing that honesty — including about risks — is both the ethical choice and the strategically sound one. 'The thing that will work is actually curing cancer,' he wrote.
It is worth noting that Amodei previously predicted AI could wipe out 50% of white-collar jobs. More recently he has described AI as a multiplier of output rather than a destroyer of jobs — a pivot that coincides with Anthropic's reported preparations for an IPO.
CEO Times Take
Amodei is right that a marketing campaign will not rebuild trust — and right that results will. Free enterprise works precisely because it ties credibility to delivery, not to promises. The more interesting signal in his post is the regulatory ask: a FINRA-style body with real teeth over frontier AI companies, including Anthropic itself. That is a significant concession from a CEO whose firm stands to be regulated, and it reflects a calculation that clear rules of the road are better for long-term capital formation than the current uncertainty.
The market will ultimately render its verdict when Anthropic files for its IPO. Until then, Amodei's candor is a better foundation than spin — and the industry would do well to follow his lead on that narrow point.
